Dynamic Machine Learning Model Selection for Cognitive Services
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users of cognitive services face sub-optimum results due to the need for complex and expensive training of machine learning models, often relying on default models not optimized for their specific applications, as creating representative training data requires subject matter expertise.
Innovation Solution
An information handling system performs multiple tests using various trained machine learning models on user data, selects the best models based on test results, and assigns the cognitive service and selected models to the user application, allowing for dynamic binding and optimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If users train machine learning models on representative data to improve accuracy, then the precision of cognitive services improves, but the complexity and cost of the process increases
Solution Approach 1:
The system pre-trains multiple machine learning models on diverse datasets before they are needed. When a user requests cognitive services, the pre-trained models are already available for evaluation and selection, eliminating the need for users to perform complex training procedures. This preliminary preparation resolves the contradiction by providing high-accuracy models without requiring users to undertake complex training processes.
Solution Approach 2:
The system creates multiple copies of machine learning models trained on different datasets and domains. Instead of requiring users to train their own models, the system provides access to these pre-trained model copies, which can be evaluated and selected based on performance metrics. This copying approach maintains high accuracy while eliminating the complexity of training.
2Adaptability or versatility
If users create custom training data for their specific application, then the adaptability of the cognitive service improves, but the time and resources required increase
Solution Approach 1:
The system performs preliminary training of machine learning models on various domains and applications before users need them. Multiple models are pre-trained on different types of data (financial, healthcare, travel, etc.), so when users need cognitive services for their specific applications, suitable pre-trained models are already available for immediate evaluation and deployment. This eliminates the time-consuming data creation and training process while maintaining application-specific adaptability.
Solution Approach 2:
The system creates machine learning models with universal applicability across multiple domains and use cases. By training models on diverse, comprehensive datasets that cover various industries and applications, the system enables a single model or set of models to serve multiple purposes. Users can select from these versatile pre-trained models without needing to create application-specific training data, thus achieving adaptability without time investment.
3Measurement precision
If multiple machine learning models are tested and selected, then the precision of the cognitive service improves, but the complexity of the system increases
Solution Approach 1:
The system implements automated feedback mechanisms that evaluate multiple pre-trained machine learning models based on performance metrics, accuracy measures, and application-specific criteria. The evaluation process automatically compares model outputs, analyzes performance differences, and provides feedback to select the optimal model. This automated feedback system resolves the contradiction by enabling precise model selection without requiring complex manual evaluation processes.
Solution Approach 2:
The system performs self-service model evaluation and selection, automatically testing multiple pre-trained models against user data and selecting the best-performing model without requiring user intervention in the complex selection process. The system independently handles model comparison, performance assessment, and selection, thereby improving result precision while keeping the user-facing complexity low. Users simply request cognitive services, and the system autonomously manages the multi-model evaluation and selection.
Data Source
AI summary
An approach is provided in which an information handling system performs multiple tests using a cognitive service and multiple trained machine learning models on user data corresponding to a user application. For each of the multiple tests, a different one of the trained machine learning models is utilized. The information handling system generates results from the tests and then selects at least one of the trained machine learning models based on the test results. In turn, the information handling system assigns the cognitive service and the selected trained machine learning models to the user application.